Authors: Jennifer Williams, Karla Pizzi, Paul-Gauthier Noe, Sneha Das
Abstract:
Most recent speech privacy efforts have focused on anonymizing acoustic speaker attributes but there has not been as much research into protecting information from speech content. We introduce a toy problem that explores an emerging type of privacy called "content masking" which conceals selected words and phrases in speech. In our efforts to define this problem space, we evaluate an introductory baseline masking technique based on modifying sequences of discrete phone representations (phone codes) produced from a pre-trained vector-quantized variational autoencoder (VQ-VAE) and re-synthesized using WaveRNN. We investigate three different masking locations and three types of masking strategies: noise substitution, word deletion, and phone sequence reversal. Our work attempts to characterize how masking affects two downstream tasks: automatic speech recognition (ASR) and automatic speaker verification (ASV). We observe how the different masks types and locations impact these downstream tasks and discuss how these issues may influence privacy goals.
Male speaker (mid-sentence masking)
p260_003 (male): Six spoons of fresh snow peas, five thick slabs of blue cheese, and maybe a snack for her brother Bob.
Original / natural
VQ-VAE / synthetic
Noise Removal Reversal
Masked Natural
Masked VQ-VAE
Female speaker (mid-sentence masking)
p294_003 (female): Six spoons of fresh snow peas, five thick slabs of blue cheese, and maybe a snack for her brother Bob.
Original / natural
VQ-VAE / synthetic
Noise Removal Reversal
Masked Natural
Masked VQ-VAE